- HyperGrid Networks
The Primary HyperGrid Network consists of many of the largest OpenSim grids as well as many smaller, personal grids. The destinations on this Network are running the more recent OpenSim revisions providing the very latest features that OpenSim has to offer.
OpenSim Core Bugs
MantisBT - Issues - (rdcdev)
Updated: 5 sec ago
In commit r/24332 (2014-02-04) the extra connector and [BakedTextureService] have been added to Robust.ini.example but not to Robust.HG.ini.example.<br /> <br /> Also, there is a missing " on the end of the BakedTextureService line in Robust.ini.example<br /> <br /> ;; Uncomment to provide bakes caching<br /> ;BakedTextureService = "8003/OpenSim.Server.Handlers.dll:XBakesConnector
Some places in code mask Environment.TickCount by 7FFFFFFF and some places mask by 3FFFFFFF.<br /> <br /> Also, some code tries to find an elapsed time by doing:<br /> int elapsed = currentTickCount & mask - previousTickCount;<br /> <br /> ...which is broken, as when:<br /> currentTickCount = 0x80000000<br /> previousTickCOunt = 0x7FFFFFFF,<br /> it yields 0x80000001<br /> <br /> ...it should be:<br /> int elapsed = (currentTickCount - previousTickCount) & mask;<br /> which yields the correct value of 1 for the above example<br /> <br /> However, it shouldn't be necessary to mask the tick count at all. The previous statement can be simply:<br /> int elapsed = currentTickCount - previousTickCount;<br /> <br /> It works even if none, either or both of current or previous tick count are negative. It breaks if the time difference is >= 2**31 ms, but so do any of the masked formulae. And masking by 3FFFFFFF breaks if the difference is >= 2**30 ms.
The problem occurs when you see the content of a variable.<br /> <br /> For example, a variable that contains the color white in RGB format is displayed like this: <1,00000,1,00000,1,00000><br /> because the decimal separator used in Italy is the comma (',') and not the dot ('.')<br /> <br /> I tried to solve the problem by modifying the source code of OpenSim.<br /> The git-diff's file is attached.
If you download <a href="http://opensimulator.org/dist/opensim-0.8.tar.gz">http://opensimulator.org/dist/opensim-0.8.tar.gz</a> [<a href="http://opensimulator.org/dist/opensim-0.8.tar.gz" target="_blank">^</a>]<br /> <br /> and go into the bin folder, you will notice a folder called j2kDecodeCashe which should not be there until the file is executed for the first time.
Use 2 Avatars starting off in the same region, in an area that has neighboring regions, cross one of the avatars over the border and send a teleport lure to the avatar still remaining in the originating region, the teleport stays hung, with no cancel button, usually it times out 2-4 minutes later, no matter how long you wait this teleport lure will never complete. In my testing the 2 regions are not in the same simulator, i have no way to test this scenario.
If you walk in a region and right click your mouse on the terrain while walking and then try to stop walking you will find that you are still walking and the region is un-responsive. Looking at the queues, the task output queue has been filled up and takes several seconds to empty.<br /> <br /> This is not just walking but also gets in the way of editing as somehow the right click causes the lag.<br /> <br /> This happens with old and new viewers (tried Phoenix even).
I would like to let avatars pick up an object when they click on it.<br /> I tried using osForceAttachToAvatar or llAttachToAvatar under touch event.<br /> After clicking on the object, I can see that the inventory showing that the object is wearing on my avatar but it is not visible actually.<br /> <br /> To solve this problem I have to edit the object / wear again after detached it in the inventory. And sometimes only my only viewer can see the change and other users cannot see the update of wearing object.<br /> <br /> * Is my approach incorrect?
Offline Hypergrid friend generate .NET errors on console,<br /> instead of just ignore the problem.<br /> <br /> OpenSim.Region.CoreModules.Framework.UserManagement.UserManagementModule [USER MANAGEMENT MODULE]: GetServerURLs call failed <br /> System.Net.WebException: Unable to connect to the remote server ---> System.Net.Sockets.SocketException: No connection could be made because the target machine actively refused it <Server ip>:9000<br /> at System.Net.Sockets.Socket.DoConnect(EndPoint endPointSnapshot, SocketAddress socketAddress)<br /> at System.Net.ServicePoint.ConnectSocketInternal(Boolean connectFailure, Socket s4, Socket s6, Socket& socket, IPAddress& address, ConnectSocketState state, IAsyncResult asyncResult, Exception& exception)<br /> --- End of inner exception stack trace ---<br /> at System.Net.HttpWebRequest.GetRequestStream(TransportContext& context)<br /> at System.Net.HttpWebRequest.GetRequestStream()<br /> at Nwc.XmlRpc.XmlRpcRequest.Send(String url, Int32 timeout)<br /> at OpenSim.Services.Connectors.Hypergrid.UserAgentServiceConnector.CallServer(String methodName, Hashtable hash) in p:\opensim-0.8-source\OpenSim\Services\Connectors\Hypergrid\UserAgentServiceConnector.cs:line 197<br /> at OpenSim.Services.Connectors.Hypergrid.UserAgentServiceConnector.GetServerURLs(UUID userID) in p:\opensim-0.8-source\OpenSim\Services\Connectors\Hypergrid\UserAgentServiceConnector.cs:line 498<br /> at OpenSim.Region.CoreModules.Framework.UserManagement.UserManagementModule.GetUserServerURL(UUID userID, String serverType) in p:\opensim-0.8-source\OpenSim\Region\CoreModules\Framework\UserManagement\UserManagementModule.cs:line 479
Scripts with a statement like<br /> <br /> (status += 32 * !(status & 32));<br /> <br /> do not compile. A statement like this comes from the expansion of a preprocessor macro. The macro is called 'SetStatus' in this case. I need it in my scripts --- which run through cpp --- for status handling.<br /> <br /> The brackets around the statement are not required in above example. They are required because the macro can be used in an expression like<br /> <br /> var1 = var2 * SetStatus(stSOMESTATUS) + var3;<br /> <br /> The same goes for other macros that expand to assignments, like replacing elements of a list, for example. The result of an assignment can be used itself as a value, like<br /> <br /> var1 = var2 = var3 = 25;<br /> <br /> Using macros introduces brackets which are not needed in all instances. However:<br /> <br /> + these brackets don't hurt anything<br /> + the scripts compile just fine in second life<br /> <br /> The syntax check that prevents scripts as described from compiling breaks compatibility with sl for no advantage at all. It should be removed, or changed so that the scripts compile.<br /> <br /> It breaks pretty much all my scritps.
The attached patch adds the user's username to the data returned from the get_user_info XML-RPC call to the User Agent service.<br /> <br /> This will be helpful to users developing web applications to be able to do both key2name and name2key.<br /> <br /> The name2key functionality is part of the get_uuid XMP-RPC call.<br /> <br /> <a href="http://opensimulator.org/pipermail/opensim-dev/2013-December/024480.html">http://opensimulator.org/pipermail/opensim-dev/2013-December/024480.html</a> [<a href="http://opensimulator.org/pipermail/opensim-dev/2013-December/024480.html" target="_blank">^</a>]
Prim can not be scaled (resized ) manually if the the permissions on the prim is enabled for copy and modify for the next owner and then create a script in it with next owner permissions enabled for copy only. Next person will not be able to scale it by going into edit window.
When clicking the avatar to look around in 3rd person, you cannot walk properly while the view is tilted.
In BulletSim the object behaves as Phantom. So it is impossible, to create, as example, an elevator.<br /> <br /> <br /> In ODE same problem, if platform, on that the Avatar stands, is very thin.<br /> <br /> Is it thick enough, the avatar is pushed sidewards but stays on the platform.<br /> <br /> <br /> In SecondLife the platform always move the avatar as expected. The Avatar stays at<br /> his place. At best he plays the walking-animation.<br /> <br /> Tested today, 2.July 2014 at SandboxPlaza I and II
When uploading meshes, the viewer has an option to include textures used with the model. Currently OpenSim correctly creates mesh and texture assets, and object inventory item for the whole upload, but does not create inventory items for uploaded textures.<br /> <br /> This behaviour diverges from SL, which creates inventory items for every uploaded texture as well. This would also be desirable result of the mesh upload in OpenSim too for several reasons: people should be able to use the same textures that have been upload with mesh, without need to re-upload them. Some export tools require the presence of the original texture in the inventory to check for the rights, etc.<br /> <br /> NewFileAgentInventory cap invoked by the viewer sends texture_folder_id UUID field where it wants textures to be uploaded.<br /> <br /> SL behaviour is to create a folder inside textures folder specified by that field with the same name as uploaded object. So when uploading object "chair 1" with 3 textures, SL will create "chair 1" folder under textures folder and put 3 inventory items inside with names:<br /> <br /> "chair 1 - Texture 1"<br /> "chair 1 - Texture 2"<br /> "chair 1 - Texture 3"<br /> <br /> It would probably be the best to implement this same behavior in opensim.
0007240: [PATCH] Include option for AutoBackupModule to remove backups older than a given number of days
I have made some changes to AutoBackupModule again. While it is possible to run a script after the backup process to clear old backups, i thought it would be easier to add the function to the module. The patch included will... <br /> <br /> 1) Add a new config variable AutoBackupKeepFilesForDays where the number of days to keep backups is specified (0 the default value will disable this option).<br /> <br /> 2) After each backup, it will remove all oars in the backup directory that are older than the specified number of days.<br /> <br /> One point to note is that if the user set the bin directory as the backup location, it's obviously important that only oar files are removed which this patch does check. Although there is no way to know if an oar was created by auto-backup and should be deleted or if its manually created or uploaded? I would expect people will specify a specific directory for auto-backup but should we assume this is so?
Over the last few weeks, I have noticed that avatars logging off (or teleporting back to a home grid) from a region on the Openvue grid (on latest 0.8.1 dev e.g. r/24902) are removed from the region they are on immediately, and they seem to be removed from immediately adjacent regions too, BUT... they seem to fail to be removed until after a 60,000ms delay from another regions they are established as "child" agents on the same grid.<br /> <br /> In this example case the region where they fail to be removed is TWO 256x256 standard regions) away, not adjacent. The log off region is at 9051,9051 and the "child" region that the avatar is removed from after 60,000ms is at 9050,9049. BOTH regions are on the same OpenSim.exe instance, though I am not sure thatis a factor.<br /> <br /> No var or mega regions are in use on the Openvue grid involved. All regions are standard 256x256 regions.<br /> <br /> Could this relate to changes in adjacent region code handling intended to allow for varregions even though var regions are not in use, or perhaps relate to max view distances being used to establish child region relationships. I do often use 512m distance settings in viewers, and others might when visiting over HG.<br /> <br /> Sample of the problem... <br /> <br /> 03:56:47 - [CLIENT]: Close has been called for lookatme.once @hg.osgrid.org attached to scene Vue-Port<br /> 03:56:47 - [SCENE]: Removing child agent lookatme.once @hg.osgrid.org df3ac0c6-e294-4426-a447-3b29a4341d0c from Vue-Port<br /> 03:56:47 - [CAPS]: Remove caps for agent df3ac0c6-e294-4426-a447-3b29a4341d0c in region Vue-Port<br /> 03:56:47 - [ENTITY TRANSFER STATE MACHINE]: Agent df3ac0c6-e294-4426-a447-3b29a4341d0c cleared from transit in Vue-Port<br /> 04:53:31 - [LLUDPSERVER]: No packets received from child agent of <a href="mailto:firstname.lastname@example.org">email@example.com</a> for 60000ms in Sandbox. Disconnecting.<br /> 04:53:31 - [CLIENT]: Close has been called for lookatme.once @hg.osgrid.org attached to scene Sandbox<br /> 04:53:31 - [SCENE]: Removing child agent lookatme.once @hg.osgrid.org df3ac0c6-e294-4426-a447-3b29a4341d0c from Sandbox<br /> 04:53:31 - [CAPS]: Remove caps for agent df3ac0c6-e294-4426-a447-3b29a4341d0c in region Sandbox
The inability to prevent HG users to join local groups allows incoming HG users to join groups, start group chats, and grief the user community on grids.
If the grid is unavailable when getting information about a HG avatar in your friends list, the first message on the console seems appropriate<br /> <br /> 09:07:48 - [USER AGENT CONNECTOR]: get_server_urls call to <a href="http://vibe.bio-se.info:9000">http://vibe.bio-se.info:9000</a> [<a href="http://vibe.bio-se.info:9000" target="_blank">^</a>] failed: The operation has timed out<br /> <br /> But then a traceback style exception occurs. Should this be supressed since the proper log message has already been given?<br /> <br /> 09:07:48 - [USER MANAGEMENT MODULE]: GetServerURLs call failed System.Net.WebException: The operation has timed out<br /> at System.Net.HttpWebRequest.GetResponse()<br /> at Nwc.XmlRpc.XmlRpcRequest.Send(String url, Int32 timeout)<br /> at OpenSim.Services.Connectors.Hypergrid.UserAgentServiceConnector.CallServer(String methodName, Hashtable hash) in d:\Temp\opensim-fe6dab\OpenSim\Services\Connectors\Hypergrid\UserAgentServiceConnector.cs:line 197<br /> at OpenSim.Services.Connectors.Hypergrid.UserAgentServiceConnector.GetServerURLs(UUID userID) in d:\Temp\opensim-fe6dab7\OpenSim\Services\Connectors\Hypergrid\UserAgentServiceConnector.cs:line 498<br /> at OpenSim.Region.CoreModules.Framework.UserManagement.UserManagementModule.GetUserServerURL(UUID userID, String serverType) in d:\Temp\opensim-fe6dab7\OpenSim\Region\CoreModules\Framework\UserManagement\UserManagementModule.cs:line 479
Opensim 0.8 distribution from the website, fresh install. Seems to be sending textures to Imprudence Viewer (Not Mesh Enabled) - however firestorm is not receiving any texture information at all. Neither is Kokua (the new version of imprudence)<br /> <br /> Things i've done:<br /> Installed FirestormOS on a Computer that has never had FirestormOS installed and received the same result (no textures) to eliminate the possibility of a casheing issue.<br /> I've cleared Cashe and done a fresh install on those systems that have had a problem in the past.
OpenSim will not start with bulletsim enabled:<br /> <br /> ...<br /> 22:39:44 - [BULLETS SCENE]: Selected bullet engine bulletunmanaged -> BulletUnmanaged/<br /> 22:39:44 - [APPLICATION]:<br /> APPLICATION EXCEPTION DETECTED: System.UnhandledExceptionEventArgs<br /> <br /> Exception: System.DllNotFoundException: lib64/libBulletSim.so<br /> at (wrapper managed-to-native) OpenSim.Region.Physics.BulletSPlugin.BSAPIUnman/BSAPICPP:Initialize2 (OpenMetaverse.Vector3,intptr,int,intptr,int,intptr,OpenSim.Region.Physics.BulletSPlugin.BSAPIUnman/BSAPICPP/DebugLogCallback)<br /> ...<br /> <br /> On further investigation both lib32/libBulletSim.so and lib64/libBulletSim.so are reported as 'not a dynamic executable' by ldd.<br /> <br /> I've tested this on two servers, both running centos 5 x86_64 with mono 2.10.2, one is an openvz VM the other a VMware VM, and get the same result. Other .so files work ok with ldd and ODE physics works fine.